Valuation of Asian Option Qi An Jingjing Guo

CONTENT Asian option Pricing Monte Carlo simulation Conclusion

ASIAN OPTION Definition of Asian option always emphasizes the gist that the payoff depends on the average price of the underlying asset over a certain period of time as opposed to at maturity. That is why Asian option could also be recognized as average value option. This type of option contract emerges and gets developed because it tends to cost less than regular American and European options. It is also known that an Asian option can protect an investor from the volatility risk that comes with the market.

TWO BASIC FORMS Average Price Option Average Strike Option.

THREE SPECIFIED TERMS 1.The averaging period. E.g. the last three month, the entire term of option etc. 2.The sampling frequency. E.g. daily, weekly, annually etc. 3.The averaging method.

AVREAGING METHOD ① simple arithmetic average ② geometric average ③ weighted average i.arithmetic weighted average ii.geometric weighted average where denotes the weight.

PRICING Three main approaches exist to pricing Asian options. 1.European-style options based on geometric averages can be priced by adapting the analytical models. The reason is if underlying price is assumed to follow log normal distribution, then its geometric average is also log normal distributed. 2.The solution above does not apply when it is based on arithmetic averages, because the arithmetic average will not be log normal distributed even if the underlying price is. It is only possible to make approximation. 3.Monte Carlo simulation could be used no matter what its style or averaging method is. As this simulation requires intensive calculation it is mostly performed by computer program.

MONTE CARLO SIMULATION Monte Carlo simulation produces results by constructing a stochastic model, where the aim solution is the expectation of the model, and then use a large number of sample results to approximate the aim solution. The law of large numbers ensures that sample mean converges to the population mean almost surely as. That is the reason why the sample mean is used as an estimate of population mean for large samples. Pricing with Monte Carlo method is assumed to be in risk-neutral world. Firstly to generate a path of underlying price stochastically, then compute the expectation of returns, at last discount it with risk- free interest rate.

MONTE CARLO SIMULATION To value an Asian relative option, which performs on two or more underlying assets, and the payoff is determined by the return of one or more assets relative the return of some other assets. To make it simpler the common case of two underlying assets is taken. The price and payoff are given by where denotes the weight for each return. In this simple case is 1 and is -1.

MONTE CARLO SIMULATION The Asian sum is defined by based on simple arithmetic average. Then it gives us In our case there are two variables and both satisfy

MONTE CARLO SIMULATION where r is risk-free interest rate, σ is the standard deviation of return on the asset price, is the standard Brownian motion under Q. So the arithmetic average price of Asian option is Because the effective unbiased estimation of E is sample mean, we can rewrite the equation as following,

function [Price,ConInterval] = aapapricemc (CurrentPrice,StrikePrice,... IntRate,TimeToMaturity,StdDeviation,CallOrPut,NumOfSteps,NumOfPaths) % AAPAPRICEMC pricing arithmetic average price of Asian option with % Monte-Carlo Simulation % PARAMETERS: % Price is the option price calculated with Monte-Carlo Simulation % ConInterval is the 95% confidence interval for the estimation of the price % CurrentPrice is the current price of the underlying asset % StrikePrice is the strike price of the option % IntRate is the risk-free interest rate % TimeToMaturity is the time to maturity % StdDeviation is the standard deviation of the return on the price of the % underlying asset % CallOrPut : if the option is a call, CallOrPut is 1; % if the option is a put, CallOrPut is 0; % in default, CallOrPut is 1; if (CallOrPut~=0)&&(CallOrPut~=1) error(sprintf('CallOrPut should be 1 or 0.')) end Paths = pricepaths(CurrentPrice,IntRate,TimeToMaturity,StdDeviation,... NumOfSteps,NumOfPaths); if CallOrPut PayOff = max(mean(Paths(:,2:(NumOfSteps+1)),2)-StrikePrice,0); else PayOff = max(StrikePrice-mean(Paths(:,2:(NumOfSteps+1)),2),0); end [Price,VarOfPrice,ConInterval]=normfit(exp(-IntRate*TimeToMaturity)*PayOff) end

Now we have our program for the calculation, then we assume some datum to calculate the price and analyze the properties. Then we substitute the datum to the function, we can get the results quickly,

It’s easy to see that, with the increase of the number of simulations, the length of confidence intervals become smaller and smaller, which means the results of simulations are more and more accurate. So we could obtain an option price very close to its real price. In option 2, 3 and 4, we changed some values of the parameters so that we have the chance to analyze the effects caused by those parameters roughly. Since we get different prices by using the exactly same values in option 1, we couldn’t get some general rule about how the other parameters affect the price by a single result. We need to do more calculations. The reason is every time we use this function in the MATLAB, the system will not create the exactly same random samples. But we can prove we can get more accurate price with Monte- Carlo simulation as the number of random samples close to infinity.

CONCLUSION Asian Options are commonly traded on currencies and commodity products which have low trading volumes. They are options that ensure to its buyer an average return at a lesser risk. Monte-Carlo simulation is a very useful tool when pricing the Asian option... Also when we try to get some solutions, we need to control all the variables and summarize the general results based on a vast number of trials.

THANKS FOR WATCHING